多移动机器人协调规划算法研究

发布时间：2018-08-12 10:58

【摘要】：通过协调控制,多个移动机器人可以实现单个移动机器人无法或难以完成的任务：搜索危险环境、运输救援物资、或围捕目标等。为了完成上述任务,多移动机器人的路径规划、编队规划、区域覆盖空洞检测与修复、以及区域中发生事件的检测及覆盖都是需要解决的关键问题。本论文主要针对多移动机器人协调控制中上述几方面关键问题展开研究,提出一系列解决上述问题的算法,研究内容主要包括以下四个方面。(1)提出了适用于多移动机器人路径规划的多克隆人工免疫网络算法在复杂、动态变化的未知环境中,根据移动机器人携带传感器获得的环境信息及机器人自身状态信息,快速、有效地规划一条安全无碰撞路径,是多移动机器人顺利完成任务的前提。动态环境中,移动障碍和移动目标的运动方程难以获得,未知环境的数学模型难以建立,如何根据障碍和目标的实时位置规划安全路径是研究的一个难点。针对存在多个静止障碍的未知环境中的单移动机器人路径规划问题,提出一种多克隆人工免疫网络算法。该算法将传感器获得的环境信息定义为抗原,将机器人可能的运动方向定义为抗体,根据生物免疫机理、抗原与抗体之间的相互作用,计算并选择出最优运动方向角。针对免疫解空间中的未成熟收敛问题,采用多克隆操作,通过增加抗体多样性解决。另外,该算法也解决了人工势场法存在的局部极小问题。多克隆人工免疫网络算法可以成功避开环境中存在的多个静止障碍,以最优路径成功到达目标位置。仿真结果及分析说明了算法的有效性,与其余算法的比较说明了算法的优越性。针对存在多个移动障碍、移动目标的未知环境中的多移动机器人动态路径规划问题,提出一种改进多克隆人工免疫网络算法。该算法在多克隆人工免疫网络算法的基础上,考虑了其余机器人、移动障碍及移动目标的影响,重新定义了抗原、抗体相互作用模型。另外,引入记忆单元,将机器人遇到过的环境信息、及针对该环境做出特异性响应的抗体存储。该算法可以根据移动障碍及移动目标的实时位置,重新选择出最优运动方向角。记忆能力不仅能够增加特异性反应被选择的概率,而且可以减少动态路径规划的响应时间。仿真结果及分析说明了算法的有效性,与其余算法的比较说明了算法的优越性。(2)提出了适用于多移动机器人编队规划的多克隆人工免疫网络算法实际应用中,为了充分获取环境信息、增强抵御外界入侵的能力、提高任务完成效率,需要对多个移动机器人进行编队。在复杂环境中,根据多移动机器人所配置传感器获得的环境信息及机器人自身状态信息,如何形成合适队形并维持、切换不同队形,同时有效避障,是多移动机器人编队研究的重点问题。针对存在多个障碍的未知环境中多移动机器人动态编队问题,采用改进多克隆人工免疫网络算法解决。根据领航机器人、跟随机器人间的期望距离和期望角度计算得到期望队形,改进多克隆人工免疫网络算法保证多移动机器人系统形成、并维持期望队形运动,同时避开障碍。图论理论的引入,实现了两种同构队形之间的平滑切换,并用于机器人避障。环境变化时,机器人角色及期望队形可以随之改变,实现动态编队。仿真结果及分析说明了算法的有效性。针对跟随机器人避障以及队形形成、队形维持问题,从不同角度、采用不同力法计算跟随机器人的运动方向角与线速度。跟随机器人的运动方向角采用多克隆人工免疫网络算法计算,不仅可以快速趋于领航机器人的方向角,而且可以成功避开障碍。跟随机器人的线速度采用位置跟踪控制法计算,可以保证跟随机器人的位置误差快速趋于零,形成期望队形。另外,根据李雅普诺夫理论证明了整个编队系统的渐进稳定性。利用仿真和MobileSim实验对算法进行了验证、比较。(3)提出了适用于移动传感器网络覆盖空洞检测与修复的sub-Voronoi单元面积法移动传感器网络中的区域覆盖具有重要的现实意义及广泛的潜在实际应用：室内外清扫、扫雷排雷、事故现场搜救、农田播种等。移动机器人可以视为传感器网络中的移动传感器节点,所有适用于传感器网络中移动节点区域覆盖的算法均适用于移动机器人区域覆盖。覆盖空洞的出现会导致监测信息感知不完整,对网络的连通性造成影响,甚至导致网络整体失效,严重影响传感器网络的性能和服务质量。在有限节点数量的传感器网络中,如何检测并修复覆盖空洞,提高区域覆盖率及节点覆盖效率,改善移动传感器网络性能,是移动传感器网络区域覆盖的研究重点。为了实现移动传感器网络区域覆盖,针对区域中出现的覆盖空洞,提出一种基于sub-Voronoi单元面积法的覆盖空洞检测与修复算法。该算法将每个Voronoi单元划分为多个sub-Voronoi单元,根据sub-Voronoi单元与节点感知圆之间的几何关系,计算每个sub-Voronoi单元内覆盖空洞的面积,进而判断是否存在覆盖空洞。为了完成覆盖空洞修复,移动节点朝向空洞面积最大的sub-Voronoi内的最优位置运动。该算法不仅可以估计出每个Voronoi单元内覆盖空洞的位置,而且可以准确计算出每一个sub-Voronoi单元内覆盖空洞的面积,同时还可以快速最大化区域覆盖率、节点覆盖效率,实现空洞修复,最大化网络利用率。利用仿真对算法进行验证、比较。(4)提出了适用于多移动机器人系统的多事件动态k-覆盖规划算法多移动机器人监测区域中可能会发生各种事件,如何以最少的能量消耗完成区域覆盖、及时检测到事件、并对事件进行覆盖,是多移动机器人区域覆盖的另一个研究重点。针对多移动机器人系统中的多事件动态k-覆盖规划问题展开研究,将其分为两个子问题：移动机器人均匀分布、节点选择。首先,为了有效与静态节点进行通信、尽可能覆盖整个区域,随机分布的稀疏移动机器人需要尽可能实现均匀部署。提出WSVHG、WSVHI两种方法实现移动机器人均匀部署。然后,在检测到事件时,采用类博弈论法选择出k个要对事件进行覆盖的移动机器人,类博弈论法考虑了节点的剩余能量、运动能量、通信能量。该算法保证机器人可以以较高的覆盖率完成区域覆盖,并且可以以较少的能量消耗完成多事件的k-覆盖。仿真验证及算法比较分析说明了算法的有效性。
[Abstract]:Through coordinated control, multiple mobile robots can accomplish tasks that a single mobile robot cannot or cannot accomplish: searching for dangerous environments, transporting rescue materials, or enclosing and catching targets. Detection and coverage are the key problems to be solved. In this paper, a series of algorithms are proposed to solve the above-mentioned problems in the coordinated control of multi-mobile robots. The research contents mainly include the following four aspects. (1) A multi-clonal artificial robot for path planning of multi-mobile robots is proposed. Immune network algorithm can quickly and effectively plan a safe collision-free path in complex and dynamic unknown environment according to the environment information obtained by the mobile robot carrying sensors and the state information of the robot itself. It is difficult to obtain the equation and establish the mathematical model of the unknown environment. How to plan the safe path according to the real-time location of obstacles and targets is a difficult problem in the research. The obtained environmental information is defined as antigen, and the possible movement direction of robot is defined as antibody. According to the biological immune mechanism and the interaction between antigen and antibody, the optimal movement direction angle is calculated and selected. The algorithm also solves the problem of local minimization in the artificial potential field method. The polyclonal artificial immune network algorithm can successfully avoid many static obstacles in the open environment and reach the target position in the optimal path. An improved polyclonal artificial immune network algorithm is proposed to solve the dynamic path planning problem of multiple mobile robots in unknown environments with multiple moving obstacles and moving targets. In addition, a memory unit is introduced to store the environment information that the robot has encountered and the antibodies that respond specifically to the environment. The algorithm can re-select the optimal direction of motion according to the moving obstacle and the real-time location of the moving target. Memory ability can not only increase the specific response but also be selected. The simulation results and analysis show the effectiveness of the algorithm and the superiority of the algorithm compared with other algorithms. (2) A polyclonal artificial immune network algorithm suitable for multi-mobile robot formation planning is proposed in practical application, in order to obtain sufficient environmental information. In complex environment, how to form a suitable formation and maintain, switch different formation and avoid obstacles effectively according to the environment information and the state information of the robot itself from the sensors configured by the multi-mobile robot is a multi-mobile problem. An improved polyclonal artificial immune network algorithm is used to solve the dynamic formation problem of multiple mobile robots in unknown environments with multiple obstacles. The method guarantees the formation of the multi-mobile robot system, maintains the desired formation motion and avoids obstacles. By introducing the graph theory, the smooth switching between the two isomorphic formation is realized and is used to avoid obstacles. When the environment changes, the robot roles and the expected formation can be changed accordingly, and the dynamic formation can be realized. Aiming at the obstacle avoidance, formation formation formation and formation maintenance of the following random robot, different force methods are used to calculate the direction angle and linear velocity of the following random robot from different angles. The position tracking control method is used to calculate the linear velocity of the random robot, which ensures that the position error of the random robot quickly approaches zero and the desired formation is formed. In addition, the asymptotic stability of the whole formation system is proved according to Lyapunov theory. (3) SubVoronoi cell area method for mobile sensor network coverage void detection and repair has important practical significance and wide potential applications: indoor and outdoor cleaning, mine clearance, accident scene search and rescue, farmland sowing and so on. As mobile sensor nodes in sensor networks, all the algorithms for coverage of mobile nodes in sensor networks are applicable to mobile robot coverage. The presence of coverage holes will lead to incomplete perception of monitoring information, affect the connectivity of the network, and even lead to the overall failure of the network, seriously affecting the sensor. Network performance and quality of service. In a limited number of sensor networks, how to detect and repair coverage holes, improve area coverage and node coverage efficiency, and improve the performance of mobile sensor networks is the focus of mobile sensor network area coverage. An algorithm for detecting and repairing coverage voids based on sub-Voronoi cell area method is proposed, which divides each Voronoi cell into several sub-Voronoi cells. According to the geometric relationship between sub-Voronoi cell and node perceptual circle, the coverage void area in each sub-Voronoi cell is calculated, and then it is judged as In order to complete the coverage void repair, the mobile node moves toward the optimal position in the sub-Voronoi cell with the largest void area. The algorithm can not only estimate the location of the void coverage in each Voronoi cell, but also accurately calculate the area of the void coverage in each sub-Voronoi cell. (4) A multi-event dynamic k-coverage planning algorithm for multi-mobile robot systems is proposed, which can be applied to multi-mobile robot monitoring areas. How to minimize the energy consumption is proposed. The problem of multi-event dynamic k-coverage planning in multi-mobile robot systems is studied. It is divided into two sub-problems: uniform distribution of mobile robots and node selection. In order to communicate effectively with static nodes and cover the whole area as much as possible, randomly distributed sparse mobile robots need to be deployed uniformly as possible. Two methods, WSVHG and WSVHI, are proposed to achieve uniform deployment of mobile robots. Then, when events are detected, K mobile robots are selected to cover the events by using game-like theory. The algorithm guarantees that the robot can complete the area coverage with a higher coverage rate and can complete the k-coverage of multi-event with less energy consumption. Simulation results and comparative analysis show that the algorithm is effective.
【学位授予单位】：山东大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP242
，

本文编号：2178841